MAutoAlign

Overview
Phase cancellation is a constant problem when recording an instrument (or instruments) with multiple microphones. It is caused by the
relatively slow speed of sound that allows the same audio signal to be received by each mike at slightly different times depending on the
distance of each from the sound source, surrounding walls etc. When these tracks are subsequently mixed, problems often occur. Some
frequencies cancel each other out and the audio loses punch resulting in a poor recording. This phenomenon is always present in varying
degrees and it cannot be fixed physically. So the first step is to acknowledge the problem and then find a way to fix it.
The solution is to delay each of the recorded tracks by a certain number of samples and most engineers do this manually. However it is not
uncommon to have delays of around 500 samples between microphones so even with just 4 microphones this gives us more than
62,500,000,000 possibilities!

MAutoAlign solves this problem in an extremely efficient way. All you need to do is place MAutoAlign on all tracks that need to be aligned,
start playback and press the Analyze button in one of them. The plugin will then analyze all tracks and determine the correct delay for
each track after just a few seconds. MAutoAlign does not alter the tracks in any way, it only delays them and inverts the phase if necessary.

Groups
When multiple instruments are recorded, drums and guitars for example, it's unlikely that any cancellations between the drum and guitar
tracks will occur as they essentially contain completely different audio. In fact, trying to fix such a group of tracks would probably do more
harm than good. In this case all that is needed is to group the different types of tracks.

For example, if 8 microphones have been used for drums, and 2 for the lead guitar, group D could be set for all 8 MAutoAlign instances in
the drum tracks and group G for both guitar tracks. Pressing Analyse in any of the drum tracks, will cause all 8 to be analysed, while the
guitar track settings will remain unchanged. Pressing Analyse in either guitar track will analyze just those two.

Spectral phase compensation system

Since version 11.02 MAutoAlign features most likely the world's first spectral phase compensation system. The delay and phase inversion
described above technically do nothing to the actual audio, so they are completely transparent, but they also have very limited power. In a
perfect world, where audio instruments would be microscopic points in space, microphones would record perfect audio without any phase
shift and no ambience, this would be fine. Unfortunately that's not the case and the reality is that whatever you do, each track will have the
phase changed in some way. In practice this means that when you mix multiple signals, some frequencies will combine satisfactorily, others
will cancel each other out making the output sound thin and, well, ugly. And although some people try to fix that problem using allpass
filters, it's an almost impossible job. MAutoAlign can now solve that in a matter of seconds.
As part of the analysis, MAutoAlign calculates the optimal phase adjustment to minimize cancellations. You can then simply enable the
Spectral phase compensation module and it will apply the phase adjustment to the input signals. It is disabled by default as it induces
latency (hence it cannot be used live) and it may not always work.
The spectral phase compensation works when there is a stable phase difference. Hence it mostly works when there is not too much noise or
ambience, ideally for close mics. Ambience and noise are very unpredictable signals with no stability whatsoever, which makes the plugin try
to come up with at least something, but the results are usually not helpful. If that's the case, try setting up a longer analysis length, which
can make the noise and ambience cancel itself out in time making the useful signal more pronounced.

Spectral phase compensation system for mixing dry & wet signals
Imagine that you take some signal, let it be a piano, and process it using some equalizer or analog device for instance. This processing most
likely has caused some phase shifting to the signal. Now you cannot really mix the processed version with the original anymore, because
some frequencies would just cancel each other out and results would be awful, or at least not exactly ideal. The spectral phase
compensation system can once again fix this for you - it will analyze the phase difference between the 2 signals and try to either revert the
phase alteration in the processed signal, or cause a similar phase alteration in the dry signal. Either way it will hopefully minimize phase
cancellation.

This also means, that for the first time you can now equalize the tracks before aligning them and MAutoAlign will not only fix the time
differences between the tracks, but also correct the phases of each track, whether it was caused by the processing, the microphone, or
anything else on the way.

Where to place MAutoAlign in the effect chain?

MAutoAlign should be placed in the insert effects chain of your host and if you don't use any plugins causing latency, then it is only a matter
of personal preference. Making MAutoAlign the first item in any chain will usually result in a more successful analysis as this will likely be
free of other effects and be the most natural input. Making it the last item lets it fix the correct the phase alterations caused by any
processors before it, such as equalizers or delays.

Problems may arise when you use latency-inducing plugins. The plugin uses the host's timestamps to synchronize all instances,
unfortunately many hosts are rather problematic in that aspect. If you exhibit erratic behaviour, where each analysis produces in different
results despite analyzing the same portion of the audio, try disabling all latency-inducing plugins temporarily. And let your host's developer
know about the problem.

How to group tracks of an acoustic drum kit?

Acoustic drums are typically recorded using several close microphones, together with overheads and room microphones. Let's start with the
"close mics".
The question is, should you group all close microphones together? There is no definitive answer, but it is customary to try it this way first. If
it doesn't provide satisfactory results, which is often the case if multiple drums are recorded using multiple microphones, you should try
grouping the individual drum kit pieces together. In most cases, the original setting with all close mics grouped together works just fine.
Although the close mics are placed very close to the individual drums, they will often capture the sound of other drums as well. This so-
called bleed usually doesn't need to be handled specifically, unless it is too loud. Drums recorded with multiple microphones, e.g. snare top
and snare bottom, should always be aligned together.
Make sure that you use a reasonable part of the recording for analysis - and that all microphones are actually capturing the performance. If
some of the tracks are silent, it is unlikely these tracks will be aligned correctly. Be especially careful when you use MAutoAlign as the last
item in the chain for each track and you also use a gate before it. By definition, the gate will be cutting out some parts of the track.
Overheads are also quite close to the drum kit, but they capture lots of ambience too. Experience tells us that this ambience usually makes
aligning pointless, however you should try anyway. Sometimes overhead microphones are placed some distance from each other. In this
case it may be necessary to align them with each other as there may be phase collision between them and alignment would improve the
stereo image.

You can also try aligning the overheads with the close mics. Your first step is to listen to discover if they are actually colliding with the close
microphones. If so, first mix all the close microphones into a bus (group) track, let's call it "closedrums" and then align the overheads with
that track. This grouping is done because aligning overheads with the separate close microphones, can confuse the analysis as the phasing
problems are minimal and the overheads contain too much ambience. So it is usually better to work with the close microphones together,
after all these are the key for a tight mix, overheads only fill the gaps, glue it together and provide some natural ambience.
Room microphones are usually placed some distance away from the whole performance and as the name suggests, are there to capture the
room. With that much ambience it is difficult to find and fix phasing issues, but it is often worth trying. As with the overheads, you may use
MAutoAlign to fix phasing issues between multiple room microphones if they have been placed far enough away from each other. The good
news is that ambience in a way fights phase cancellations, so in most cases there actually isn't any problem in the first place.

Presets
Presets button shows a window with all available presets. A preset can be loaded from the preset window by double-clicking on it, using the
arrow buttons or by using a combination of the arrow keys and Enter on your keyboard. You can also manage the directory structure, store
new presets, replace existing ones etc. Presets are global, so a preset saved from one project, can easily be used in another.

Holding Ctrl while pressing the button loads an existing preset, selected at random.
Presets can be backed up by using either the Export button, or by saving the actual preset files, which are found in the following directories:
Windows: C:\Users\{username}\AppData\Roaming\MeldaProduction
Mac OS X: ~/Library/Application support/MeldaProduction
Exported preset files can be loaded into the plug-in's preset store using the Import button. Or the preset files themselves can be copied into
the directories named above.
Files are named based on the name of the plugin in this format: "{pluginname}presets.xml", for example: MAutopanpresets.xml or
MDynamicspresets.xml. If the directory cannot be found on your computer for some reason, you can just search for the particular file.

Left arrow
Left arrow button loads the previous preset.

Right arrow
Right arrow button loads the next preset.

Channel mode
Channel mode button shows the current processing channel mode, e.g. Left+Right (L+R) indicates the processing of left and right
channels. This is the default mode for mono and stereo audio material and effectively processes the incoming signal as expected. However
the plugin also provides additional modes, of which you may take advantage as described below. Mastering this feature will give you
unbelievable options for controlling the stereo field.
Note that this is not relevant for mono audio tracks, because the host supplies only one input and output channel.

Left (L) mode and Right (R) mode allow the plugin to process just one channel, only the left or only the right. This feature has a
number of simple uses. Equalizing only one channel allows you to fix spectral inconsistencies, when mids are lower in one channel for
example. A kind of stereo expander can be produced by equalizing each side differently. Stereo expansion could also be produced by using a
modulation effect, such as a vibrato or flanger, on one of these channels. Note however that the results would not be fully mono compatible.
Left and right channels can be processed separately with different settings, by creating two instances of the plugin in series, one set to 'L'
mode and the other to 'R' mode. The instance in 'L' mode will not touch the right channel and vice versa. This approach is perfectly safe and
is even advantageous, as both sides can be configured completely independently with both settings visible next to each other.
Mid (M) mode allows the plugin to process the so-called mid (or mono) signal. Any stereo signal can be transformed from left and right, to
mid and side, and back again, with minimal CPU usage and no loss of audio quality. The mid channel contains the mono sum (or centre),
which is the signal present in both left and right channels (in phase). The side channel contains the difference between the left and right
channels, which is the "stereo" part. In 'M mode' the plugin performs the conversion into mid and side channels, processes mid, leaves side
intact and converts the results back into the left and right channels expected by the host.
To understand what a mid signal is, consider using a simple gain feature, available in many plugins. Setting the plugin to M mode and
decreasing gain, will actually lower or attenuate the mono content and the signal will appear "wider". There must be some stereo content
present, this will not work for monophonic audio material placed in stereo tracks of course. Similarly amplifying the mono content by
increasing the gain, will make the mono content dominant and the stereo image will become "narrower".
As well as a simple gain control there are various creative uses for this channel mode.
Using a compressor on the mid channel can widen the stereo image, because in louder parts the mid part gets attenuated and the stereo
becomes more prominent. This is a good trick to make the listener focus on an instrument whenever it is louder, because a wider stereo
image makes the listener feel that the origin of the sound is closer to, or even around them.
A reverb on the mid part makes the room appear thin and distant. It is a good way to make the track wide due to the existing stereo
content, yet spacey and centered at the same time. Note that since this effect does not occur naturally, the result may sound artificial on its
own, however it may help you fit a dominant track into a mix.
An equalizer gives many possibilities - for example, the removal of frequencies that are colliding with those on another track. By processing
only the mid channel you can keep the problematic frequencies in the stereo channel. This way it is possible to actually fit both tracks into
the same part of the spectrum - one occupying the mid (centre) part of the signal, physically appearing further away from the listener, the
other occupying the side part of the signal, appearing closer to the listener.
Using various modulation effects can vary the mid signal, to make the stereo signal less correlated. This creates a wider stereo image and
makes the audio appear closer to the listener.
Side (S) mode is complementary to M mode, and allows processing of only the side (stereo) part of the signal leaving the mid intact. The
same techniques as described for M mode can also be applied here, giving the opposite results.
Using a gain control with positive gain will increase the width of the stereo image.
A compressor can attenuate the side part in louder sections making it more monophonic and centered, placing the origin a little further
away and in front of the listener.
A reverb may extend the stereo width and provide some natural space without affecting the mid content. This creates an interesting side-
effect - the reverb gets completely cancelled out when played on a monophonic device (on a mono radio for example). With stereo
processing you have much more space to place different sounds in the mix. However when the audio is played on a monophonic system it
becomes too crowded, because what was originally in two channels is now in just one and mono has a very limited capability for 2D
placement. Therefore getting rid of the reverb in mono may be advantageous, because it frees some space for other instruments.
An equalizer can amplify some frequencies in the stereo content making them more apparent and since they psycho acoustically become
closer to the listener, the listener will be focused on them. Conversely, frequencies can be removed to free space for other instruments in
stereo.
A saturator / exciter may make the stereo richer and more appealing by creating higher harmonics without affecting the mid channel,
which could otherwise become crowded.
Modulation effects can achieve the same results as in mid mode, but this will vary a lot depending on the effect and the audio material. It
can be used in a wide variety of creative ways.
Mid+Side (M+S) lets the plugin process both mid and side channels together using the same settings. In many cases there is no
difference to L+R mode, but there are exceptions.
A reverb applied in M+S mode will result in minimal changes to the width of the stereo field (unless it is true-stereo, in which case mid will
affect side and vice versa), it can be used therefore, to add depth without altering the width.
A compressor in M+S mode can be a little harder to understand. It basically stabilizes the levels of the mid and side channels. When
channel linking is disabled in the compressor, you can expect some variations in the sound field, because the compressor will attenuate the
louder channel (usually the mid), changing the stereo width depending on the audio level. When channel linking is enabled, a compressor
will usually react similarly to the L+R channel mode.
Exciters or saturators are both nonlinear processors, their outputs depend on the level of the input, so the dominant channel (usually
mid) will be saturated more. This will usually make the stereo image slightly thinner and can be used as a creative effect.
How to modify mid and side with different settings? The answer is the same as for the L and R channels. Use two instances of the
plugin one after another, one in M mode, the other in S mode. The instance in M mode will not change the side channel and vice versa.
Left+Right(neg) (L+R-) mode is the same as L+R mode, but the the right channel's phase will be inverted. This may come in handy if
the L and R channels seem out of phase. When used on a normal track, it will force the channels out of phase. This may sound like an
extreme stereo expansion, but is usually extremely fatiguing on the ears. It is also not mono compatible - on a mono device the track will
probably become almost silent. Therefore be advised to use this only if the channels are actually out of phase or if you have some creative
intent.

There are also 4 subsidiary modes: Left & zero Right (L(R0)), Right & zero Left (R(L0)), Mid & zero Side (M(S0)) and Side & zero
Mid (S(M0)). Each of these processes one channel and silences the other.
Surround mode is not related to stereo processing but lets the plugin process up to 8 channels, depending on how many the host supplies.
For VST2 plugins you have to first activate surround processing using the Activate surround item in the bottom. This is a global switch for
all MeldaProduction plugins, which configures them to report 8in-8out capabilities to the host, on loading. It is disabled by default, because
some hosts have trouble dealing with such plugins. After activation, restart your host to start using the surround capabilities of the plugins.
Deactivation is done in the same way. Please note that all input and output busses will be multi-channel, that includes side-chain for
example. For VST3/AU/AAX plugins the activation is not necessary.
First place the plugin on a surround track - a track that has more than 2 channels. Then select Surround from the plug-in's Channel Mode
menu. The plugins will regard this mode as a natural extension of 2 channel processing. For example, a compressor will process each
channel separately or measure the level by combining the levels of all of the inputs provided. Further surround processing properties, to
enable/disable each channel or adjust its level, can be accessed via the Surround settings in the menu.
Ambisonics mode provides support for the modern 3D systems (mostly cinema and VR) with up to 64 channels (ambisonics 7th order).
Support for this is still quite rare among the DAWs, so this needs to be activated in all DAWs using the Activate ambisonics item in the
bottom. This is a global switch for all MeldaProduction plugins, which configures them to report 64in-64out capabilities to the host, on
loading. After activation, restart your host to start using the ambisonics capabilities of the plugins. Deactivation is done in the same way.
Please note that all input and output busses will be multi-channel, that includes side-chain for example.
First place the plugin on an ambisonics track, supported are all orders from 1st (4 channels) to 7th (64 channels). Then select Ambisonics
from the plug-in's Channel Mode menu. Finally select the Ambisonics settings in the menu and configure the Ambisonics order and other
settings if needed. The plugins will regard this mode as a natural extension of 2 channel processing. For example, a compressor will process
each channel separately or measure the level by combining the levels of all of the inputs provided.

Panic
Panic button resets the plugin state. You can use it to force the plugin to report latency to the host again and to avoid any audio problems.
For example, some plugins, having a look-ahead feature, report the size of the look-ahead delay as latency, but it is inconvenient to do that
every time the look-ahead changes as it usually causes the playback to stop. After you tweak the latency to the correct value, just click this
button to sync the track in time with the others, minimizing phasing artifacts caused by the look-ahead delay mixing with undelayed audio
signals in your host. It may also be necessary to restart playback in your host.
Another example is if some malfunctioning plugin generates extremely high values for the input of this plugin. A potential filter may start
generating very high values as well and as a result the playback will stop. You can just click this button to reset the plugin and the playback
will start again.

Settings
Settings button shows a menu with additional settings of the plugin. Here is a brief description of the separate items.
Activate lets you activate the plugin if the drag & drop activation method does not work in your host. In this case either click this button
and browse to the licence file on your computer and select it. Or open the licence file in any text editor, copy its contents to the system
clipboard and click this button. The plugin will then perform the activation using the data in the clipboard, if possible.
There are 4 groups of settings, each section has its own detailed help information: GUI & Style enables you to pick the GUI style for the
plug-in and the main colours used for the background, the title bars of the windows and panels, the text and graphs area and the
highlighting (used for enabled buttons, sliders, knobs etc).

Advanced settings configures several processing options for the plug-in.

Dry/wet affects determines, for Multiband plug-ins, which multiband parameters are affected by the Global dry/wet control.
Smart interpolation adjusts the interpolation algorithm used when changing parameter values; the higher the setting the higher the audio
quality and the lower the chance of zippering noise, but more CPU will be used.

WWW
WWW button shows a menu with additional information about the plugin. You can check for updates, get easy access to support,
MeldaProduction web page, video tutorials, Facebook/Twitter/YouTube channels and more.

Group
Group lets you select multiple tracks to synchronize with each other. All tracks with the same group are analyzed together.

Latency
Latency lets all of the instances in the group report this latency. Since the process of aligning delays the tracks, they slightly shift in time, but
always forwards. You may therefore experience a feeling of different timing. To compensate for it, the plugin can report latency, for which
the host compensates (Plugin Delay Compensation, or PDC), which in effect moves all the audio backwards in time. It is usually necessary to
restart playback after changing this value.
Range: 0 ms to 100 ms, default 0 ms

Delay
Delay lets you manually move the computed offset. You can use this to adjust the results if you are not satisfied with them. What you do is
immediately updated in the log. You cannot have negative delay, since it is impossible without inducing latency. But note that the meters
showing better values may not mean the results are actually better. You should always trust your ears.
Range: -1024 samples to 1024 samples, default 0 samples

Analysis panel
Analysis panel contains the parameters of the standard delay and phase inversion analysis.

Zero latency during analysis

Zero latency during analysis makes all the plugins in the group report 0 latency when the analysis starts and put it back again
afterwards. This is useful for hosts, which do not report correct timing info to the plugins when a plugin with latency is used. So basically
leave this disabled and if you feel like things are not working properly, turn it on. Note that it makes a difference only if the plugin is
actually having latency, typically when spectral phase compensation is used.

Length
Length controls the length of the analysis, hence how much signal needs to be recorded for the analysis. The default should suit most
needs, but if you feel like you are processing long and steady signals, you may increase the length for example. Note that increasing the
length does not improve the accuracy of analysis and sometimes it can even make things worse.
Range: 2000 ms to 30000 ms, default 4000 ms

Maximum delay
Maximum delay defines maximum time offset to be detected. In most cases the delay between microphones is lower than 10ms and
making this value higher can make the plugin generate higher delays due repeating sequences in the audio. Please note that this
parameter is relevant only when you click the Analyse button in.
Range: 0.10 ms to 1000 ms, default 10 ms

High-pass
High-pass defines the high-pass filter cut-off frequency which may be used to remove part of the signal for the analysis, for example to
avoid drum leakage. Please note that this filter is used only for analysis and does not affect the audio signal itself.
Range: Off to 20.0 kHz, default Off

Low-pass
Low-pass defines the low-pass filter cut-off frequency which may be used to remove part of the signal for the analysis, for example to
avoid drum leakage. Please note that this filter is used only for analysis and does not affect the audio signal itself.
Range: 20.00 Hz to Off, default Off

Phase resolution
Phase resolution controls the block size used for spectral phase compensation analysis and processing. The algorithm is working in the
so-called spectral domain. This allows it to access features that are normally unavailable, however in order to do that it requires the
audio to be separated into blocks of audio. As a result, using the spectral phase compensation causes latency. This setting controls the
latency length. Additionally, the higher it is the more detail the plugin has, which usually provides higher audio quality, at the expense of
greater CPU cost and increased latency.
Range: 256 to 65536, default 4096

Noise floor
Noise floor controls the input volume considered to be the noise floor, which helps the plugin distinguish between useful signal and noise
during analysis. It has no effect on the audio or metering.
Range: silence to 0.00 dB, default -80.0 dB

Spectral phase compensation panel

Spectral phase compensation panel contains parameters of the advanced algorithm, which directly adjusts the phase of all tracks to
avoid phase cancellations. Please note that once you enable this panel, all of the instances in the group will start inducing latency.

Bypass
Bypass switch disables the spectral phase compensation without changing the latency, which is what the Enable button would do,
which makes it hard to use for comparison, since is causes a lag. Bypass is a better choice which lets you properly determine if the
phase compensation is worth keeping enabled.

Threshold
Threshold defines the minimum level to be considered relevant for the phase shifting. In most signals all frequencies are actually
present, but most of them are only noise, distortion and other artefacts. Shifting such signals wouldn't be a problem on its own, but the
more the phase is changed, the bigger the chance that something will go wrong. In percussive signals this could lead to smearing of the
initial transient for example. This parameter could be used to let the plugin consider only the prominent frequencies relevant for the
analysis. All frequencies will then be processed anyway, but this can make the resulting phase envelope much less complex, which in
effect improves the phase coherency and minimizes the chance that something goes wrong.
Range: silence to 0.00 dB, default -60.0 dB

Smoothing
Smoothing lets you smoothe the phase envelope. It could be helpful if the processing is too brutal, causing various artefacts such as
transient smearing for percussive signals. However, please note that by using this parameter you also get further away from the
theoretically ideal phase processing, so the phase cancellations can increase again.
Range: 0.00% to 100.0%, default 10.0%

Depth
Depth controls the amount of phase shift being performed. In most cases only 100% makes sense as it performs the exact phase shift
determined as ideal. However you may want to change this as an experiment to see if different values could provide better results.
Range: 0.00% to 200.0%, default 100.0%

Disable delay
detection
Disable delay detection switch makes the analysis avoid detecting the delay and inversion and focus on phase compensation only.
Normally you should leave this disabled as both things work together. However in some cases, such as mixing a dry signal with a wet
 one (with altered phase using an equalizer for example), where you know that no delay has been performed, you may try enabling this
option to see if it would result in a better analysis.

Optimize delay
Optimize delay switch enables advanced processing which optimizes both delays and phase compensation at once. It is performed
during the analysis, so to make this work, you need to have both spectral phase compensation and optimize delay enabled. It can
provide better audio quality.

Show wrapped
phase
Show wrapped phase switch changes the phase graph into a so-called wrapped one. Phase is not a classic measure, it is cyclic, meaning
that -180 degrees produces the same effect as +180 degrees. Showing the phase in such a diagram (called wrapped) can then look
rather hard to understand, since in many cases the line will be jumping from top to bottom eventually looking like a big mess. Therefore
by default the plugin shows so-called unwrapped phase, which adjusts for that and makes the graph look easier to understand for a
human.

Phase graph display

Phase graph display shows the phase compensation for this instance after processing by Threshold and other controls. The graph
basically shows how much the phase of each frequency is shifted. Note that since the phase adjustment can reach huge numbers
(which is not a bad thing due to the fact that phase is cyclic), the vertical units don't show the actual phase, but the number of half-
cycles - 1 means 180 degrees, 2 means 360 degrees. In any case these values are useful for your curiosity only.

Analyse
Analyse button starts the analysis. Start playback in your host at an averagely loud position and press this button. The processing will take a
few seconds and new settings will appear immediately. If you aren't be satisfied with the results, you can just press the button again to
proceed with another analysis. The reason is, that in most real cases there is no perfect solution and results highly depend on the actual
part of the song being played during the analysis.

Bypass all
Bypass all button bypasses all plugins in the group as opposed to the standard bypass, which overrides just the single instance. Bypassing all
lets you check how all tracks sound without the automatic alignment.

A/B
A/B button lets you switch between 2 analyses. If you switch from A to B, the plugin changes all other plugins in the same group as well.
Analysis results are then stored in the particular slot, so you when you have analyzed both, you can easily switch between them using this
button and audition which sounds better.

Log view
Log view displays calculated delay in samples, milliseconds, and the approximate distance in centimeters and inches.

INV
INV button is highlighted if the phase has been inverted. You can change the state by clicking on it.